The Evolution and Evaluation of a Whole Blood Polymerase Chain Reaction Assay for the Detection of Invasive Aspergillosis in Hematology Patients in a Routine Clinical Setting

M A J O R A R T I C L E
The Evolution and Evaluation of a Whole Blood
Polymerase Chain Reaction Assay for the Detection
of Invasive Aspergillosis in Hematology Patients
in a Routine Clinical Setting
P. Lewis White,1 Christopher J. Linton,2 Michael D. Perry,1 Elizabeth M. Johnson,2 and Rosemary A. Barnes1
1Department of Medical Microbiology and National Public Health Service Cardiff, University Hospital of Wales, Cardiff, and 2Mycology Reference
Laboratory, Health Protection Agency, Southwest Regional Laboratory, Bristol, United Kingdom
(See the editorial commentary by Donnelly on pages 487–9)
Background.
Invasive aspergillosis (IA) is associated with high mortality. Successful outcome with treatment
is linked to early diagnosis. The utility of classic diagnostic methods, however, is limited.
Methods.
To aid in the diagnosis of IA, we retrospectively assessed our diagnostic service, using real-time
polymerase chain reaction (PCR) and galactomannan sandwich enzyme-linked immunosorbent assay (ELISA).
Results.
A total of 203 patients at risk of invasive fungal infection were screened by PCR, and 116 of the
patients were also tested by ELISA. The patient group comprised 176 patients with hematological malignancy and
28 control patients with evidence of invasive candidal infection. Consensus European Organisation for Research
and Treatment of Cancer and Mycoses Study Group criteria were used to classify fungal infection, which, by
de?nition, excluded the PCR result. The PCR method was sensitive (up to 92.3% sensitivity) and speci?c (up to
94.6% speci?city) and had good agreement with the galactomannan ELISA (76.7%) and high-resolution computed
tomography scan results.
Conclusions.
A negative PCR result can be used to rule out IA and to limit the need for empirical antifungal
therapy; thus, it has a role in diagnosing IA infections, especially in combination with antigen testing. PCR-positive
cases classi?ed as “false positives” regularly re?ect the limitations of classic microbiological procedures or restricted
use of consensus clinical methods employed to classify infection.
The incidence of invasive aspergillosis (IA) is increasing
pending on the target gene, assays demonstrate high
[1]. Dif?culties with diagnosis mean that most invasive
speci?city, with ?gures of 100% [4–6], 93% [7], and
fungal infections are proven only at autopsy. Histo-
92% [8] reported. Other factors, including the ef?cacy
pathological analysis can establish diagnosis of a ?la-
of the DNA extraction [9, 10], quality of primers (dimer
mentous fungal infection but rarely can identify the
formation and/or nonspeci?c ampli?cation), type of
agent at a species level. High-resolution CT scan of the
PCR reaction (block-based PCR [11], nested PCR [12],
chest may be a useful adjunctive investigation in di-
or real-time PCR [5]), size of population studied, type
agnosis but cannot con?rm infection, and results may
of disease [13], and sample type (plasma, whole blood,
be transient [2, 3].
or bronchoalveolar lavage [BAL] [14–16]) will affect
DNA-based methods have shown potential in the
the sensitivity of the assay.
de?nitive diagnosis of invasive fungal infections. De-
The method of Williamson et al. [12] has been used
in our laboratories for investigational diagnosis. Oc-
casional problems with nonspeci?c ampli?cation hin-
Received 24 May 2005; accepted 25 September 2005; electronically published
dering the interpretation of results, plus the additional
17 January 2006.
processing time required to perform gel electrophoresis,
Reprints or correspondence: Dr. Lewis White, Dept. of Medical Microbiology
and NPHS Cardiff, University Hospital of Wales, Heath Park, Cardiff, CF14 4XN,
suggested that the assay could be improved by using
United Kingdom (Lewis.White@nphs.wales.nhs.uk).
an Aspergillus-speci?c probe and the Roche Light Cycler
Clinical Infectious Diseases
2006; 42:479–86
for the second round of nested PCR. The aim of this
? 2006 by the Infectious Diseases Society of America. All rights reserved.
1058-4838/2006/4204-0007$15.00
study was to evaluate the performance of the modi?ed
Aspergillus PCR • CID 2006:42 (15 February) • 479

---

assay in a clinical setting as an adjunct to diagnosis by accepted
ni?cant infection. Hematology patients without host factors or
European Organisation for Research and Treatment of Cancer
with an alternative cause of fever with no clinical or micro-
and Mycoses Study Group (EORTC-MSG) criteria [17].
biological features of invasive fungal infection were classi?ed
as at-risk. In addition, 28 nonhematology, critical-care patients
PATIENTS AND METHODS
with clinical and microbiological evidence of invasive candidal
infections and no evidence of IA were included as a control
Patient Group
population to test the speci?city of the assay and were classi?ed
A group of 203 patients at risk of invasive fungal infection were
as at-risk (table 1).
tested by real-time PCR over a 13-month period (November
2003–December 2004). The majority (176) were hematology
Sample Type
patients, with 133 receiving remission-induction therapy for
For PCR, twice-weekly 2-mL whole blood samples collected in
acute leukemia (68 patients) or undergoing stem cell trans-
EDTA vacutainers were requested and were stored at ?80?C
plantation (65 patients) (table 1). The mean age of patients
until testing. Clinicians requesting Aspergillus antigen tests were
was 48 years (table 1). The EORTC-MSG criteria [17] (outlined
instructed to collect clotted blood specimens twice weekly, and
in table 2) were used to classify fungal infection (table 1).
2 mL of serum was removed and was stored at ?80?C until
Stem cell transplant recipients and patients with acute mye-
testing. Samples were stored for up to 7 days before testing.
loblastic leukemia received itraconazole prophylaxis, and pa-
tients with acute lymphoblastic leukemia received ?uconazole
prophylaxis. Piperacillin/tazobactam and amikacin were used
DNA Extraction
for empirical treatment of febrile neutropenia, with lipid for-
From fungal cultures.
Cultures were grown overnight at 37?C
mulations of amphotericin or echinocandin added for refrac-
in glucose broth. The following day, fungal biomass (1–2 mL
tory fever after 96 h.
of culture) was harvested by centrifugation (at 5300 g for 10
In accordance with proposed standards of care [3], high-
min), and the supernatant was discarded. Fungal DNA was
resolution CT scans were requested by clinicians but were delayed
then extracted as described, with the exclusion of the blood
or refused on some occasions (resulting compliance, 36.9%), and
lysis steps.
bronchoscopic analysis was performed if an undiagnosed respi-
From clinical specimens.
Fungal DNA was extracted from
ratory pathogen was suspected. Nine patients had Aspergillus
blood using a modi?ed version of the semiautomated method
species cultured from respiratory specimens (7 sputum and 2
described by Loef?er et al. [9]. All manual steps of the extrac-
BAL specimens). Aspergillus fumigatus was the most common
tion procedure were performed in a class 2 laminar ?ow cabinet.
species isolated (n
After lysis of the RBCs and WBCs [18], samples were bead-
p 7), although Aspergillus niger and Aspergillus
terreus were also cultured. Contemporaneous samples for PCR
beaten with ?20 mL of acid-washed glass beads (1180 microns;
and galactomannan ELISA (BioRad) were taken at the decision
Sigma) and were pulse-centrifuged before being washed with
of the clinicians, who were advised to obtain samples twice weekly
200 mL of molecular-grade water. The 200-mL washings were
(Monday and Thursday) as part of the investigation of febrile
transferred into a MagNA Pure LC sample cartridge and were
neutropenia in high-risk patients, although sampling was not
loaded onto a MagNA Pure platform (Roche), and the extrac-
always consistent (compliance, 50%).
tion was completed using the Total NA serum, plasma, blood
The galactomannan ELISA was performed as described by
kit, and program (Roche). The DNA was eluted in a volume
the manufacturer. To comply with European Directive 98/79/
of 100 mL.
EC, only index values of ?1.5 were reported as positive, al-
In each clinical run, A. fumigatus conidia were quanti?ed
though the absolute values were recorded, and the number of
using a Fuchs Rosenthal counting chamber, and known levels
probable cases of IA were not affected by a reduction of the
of A. fumigatus (10–20 colony-forming units [cfu]/mL) were
index factor to 0.5.
spiked into (i.e., added to) whole blood samples collected in
Retrospectively, patients were assessed for evidence of bac-
EDTA vacutainers from a healthy donor, to provide a positive
terial or viral infections according to standard diagnostic pro-
extraction control. Nonspiked blood from a healthy donor was
tocols. Signi?cant pathogens isolated from blood cultures (ex-
used as a negative extraction control.
cluding single isolates of coagulase-negative Staphylococcus),
positive cytomegalovirus PCR results, or positive immuno?uo-
PCR
rescence ?ndings for respiratory viruses resulted in the reclas-
Production of standards.
By use of primers (AF4 and AR1)
si?cation of possible IA cases as “at-risk” cases. Probable cases
and methods described elsewhere [12], a 270-bp PCR prod-
were not reclassi?ed, as the evidence for probable IA retains
uct was ampli?ed from A. fumigatus DNA (clinical isolate).
its weight even if other infections are diagnosed. However, no
PCR products were cloned using the TOPO TA Cloning kit
patients with probable IA had evidence of any alternative sig-
(Invitrogen), incorporating a pCRII-TOPO vector and One
480 • CID 2006:42 (15 February) • White et al.

---

Table 1.
Data for the patients (n p
)
203 tested by PCR.
from 45 strains from 13 species of Aspergillus were compared
for homology. Sequences from representative Aspergillus species
Characteristic
Value
were compared with the sequences from 23 species from 10
Age group, years
different genera. A hydrolysis (TaqMan) probe was designed to
!10
6
hybridize to a sequence that was speci?c to the genus Asper-
11–20
15
gillus. The binding sites for the hydrolysis probe ASP28P (FAM-
21–30
9
CATTCGTGCCGGTGTACTTCCCCG-TAMRA) and primers
31–40
28
ASF1 and ADR1 on A. fumigatus (NCBI Nucleotide Database
41–50
23
accession number AJ438345) are at positions 453–486, 392–
51–60
45
?
412, and 560–575, respectively.
61
75
Roche Light Cycler assay.
Real-time PCR was performed
Not given
2
Diagnosis/condition
using the Aspergillus-speci?c primers and probes (ASF1, ADR1,
Acute myeloid leukemia
58
and ASP28P). For amplicon detection, the Light Cycler Fast
Acute lymphoblastic leukemia
10
Start DNA Master Hybridisation Probes kit was used. The PCR
Chronic myeloid leukemia
2
mix contained 1? Fast Start reaction mix, 0.75 mmol/L of each
Chronic lymphoblastic leukemia
3
primer, 0.4 mmol/L of probe, 4 mmol/L of MgCl , and DNA
2
Other leukemia
2
template, for a ?nal volume of 20 mL. PCR conditions were 1
Lymphoma
22
cycle at 95?C for 15 min followed by 60 cycles at 95?C for 5 s
Myeloma
13
and 60?C for 30 s, while data were acquired at channel F1/F2.
Allogeneic SCT and nonspeci?ed malignancy
65
In addition to extraction controls for clinical runs, PCR stan-
Nonhematological disordera
28
dards in the form of cloned PCR products (200, 20, and 5 input
Galactomannan ELISA performed (%)
116 (57)
copies) and duplicate DNA-free PCR controls (molecular-grade
No. of samples tested (range per patient [n p
]
203 )
401 (1–14)
water) were included in every run. PCRs with positive results
No. of samples with IA classi?cation
Proven/probable IA (range per patient [n
were repeated for con?rmation and were considered clinically
p
]
14 )
70 (2–14)
Possible IA (range per patient [n
signi?cant when serial positive results were generated during a
p
]
40 )
117 (2–7)
At-risk (range per patient [n p
]
149 )
214 (1–10)
single episode, although widespread use of empirical antifungal
agents was a possible confounding factor affecting positivity.
NOTE.
Data are no. of patients, unless otherwise indicated.
a Nonhematological cases included respiratory failure, esophageal rupture,
To further enhance the sensitivity of the assay, it was nec-
cystic ?brosis, diabetes, pneumonia, premature birth, renal failure, cirrhosis,
essary to perform a nested PCR. The ?rst round of the nested
or postsurgical complications following cardiac or abdominal surgery. All cases
PCR was done as described elsewhere [12], but primer con-
had clinical and microbiological evidence of invasive candidal infections (11
proven candidemia cases and 7 probable and 10 possible cases of invasive
centrations were increased to 0.4 mmol/L, and DNA template
candida infection). Proven cases were de?ned by the isolation of Candida from
was increased to 10 mL. On completion of the ?rst round, 10
sterile sites (excluding drain ?uids and urine) and by histopathological analysis.
Probable invasive Candida infections were classi?ed by the presence of risk
mL of the amplicon was used as the template for the second
factors (surgery, antibiotic use, prematurity, venous catheters, parenteral nu-
round on the Light Cycler, by use of conditions described above.
trition, and immunosuppression), with clinical signs of infection (fever unre-
sponsive to antibacterial antibiotics) and Candida colonization at multiple (12)
Controls were as described above except for 2 additional DNA-
noncontiguous anatomical sites. Furthermore, in neonates, persistent candi-
free PCR controls.
duria was considered signi?cant. Possible invasive Candida infections were
de?ned by clinical signs of infection linked to predisposing conditions, such
as necrotizing pancreatitis, anastamotic breakdown or recurrent perforation of
RESULTS
the bowel, or spontaneous esophageal rupture with no evidence of a clinically
signi?cant bacteremia [14]. SCT, stem cell transplant.
Aspergillus PCR detection limits, reproducibility, and cross-
reactivity.
The detection limit for the assay was determined
Shot Top 10F chemically competent Escherichia coli. Plasmids
for both cfu and plasmid copy number. Serial dilutions showed
were extracted (S.N.A.P. MiniPrep Kit; Invitrogen) from Luria-
that the lower limit of detection of the Light Cycler assay alone
Bertani and ampicillin broth cultures and were quanti?ed using
was ?2 input copies, although, for 100% reproducibility, 5
a GeneQuant DNA/RNA calculator (Amersham Biosciences).
input copies were necessary (?gure 1). The standard curve for
To con?rm the identity of the insert, the plasmids were screened
the Aspergillus assay is shown in ?gure 2.
with direct PCR, with use of the primers AF4 and AR1, and
The single-round Light Cycler assay could reproducibly detect
were digested with the restriction enzyme EcoRI.
10 cfu/mL spiked in blood, with a lower detection limit of !10
Oligonucleotide design.
Aspergillus-speci?c primers de-
cfu (?gure 1). The nested Aspergillus assay improved detection
scribed elsewhere [12] were used. The probe was designed to
limits to 1 input copy and !5 cfu reproducibly detected.
bind an Aspergillus-speci?c region of the 28S rRNA gene am-
Four species of Aspergillus (A. fumigatus, Aspergillus ?a-
pli?ed by the primers ASF1 and ADR1. Analogous sequences
vus, Aspergillus nidulans, and A. niger), a strain of Fusarium
Aspergillus PCR • CID 2006:42 (15 February) • 481

---

Table 2.
Criteria used to determine certainty of invasive aspergillosis (IA)
infection.
IA infection criteria
Host factors
Antibacterial refractory fever (for 196 h)
Neutropenia (neutrophil count, !500 neutrophils/mm3 for 110 days)
Fever of unknown origin
Prolonged steroid use (13 weeks)
Graft-versus-host disease
Microbiological factors
Positive result of culture of sputum, bronchoalveolar lavage, or sinus specimen
Positive microscopic analysis result
Positive galactomannan antigen test result
Clinical factors
Major
CT ?ndings (halo, air crescent)
Nodular skin rash
Minor
New pleural in?ltrate (on chest radiograph)
Pleuritic chest pain
Shortness of breath
Cough
Hemoptysis
Pleural rub
NOTE.
Proven IA is de?ned by histological evidence with a concomitant positive culture
result. Probable IA is de?ned by a host factor plus a microbiological factor plus 1 major (or 2
minor) clinical factors. Possible IA is de?ned by a host factor plus a microbiological factor or
1 major (or 2 minor) clinical factors. Criteria from [17].
solani, and 15 species of yeast (4 genera) were tested for cross-
samples were excluded from statistical analysis, although con-
reactivity. All Aspergillus species tested were detected by the
sidering them false-positive results had little impact on the assay
assay, whereas the 12 Candida species (Candida albicans, Can-
performance. One patient classi?ed as having probable IA was
dida dubliniensis, Candida famata, Candida glabrata, Candida
excluded from analysis because samples were sent 2 weeks after
guilliermondii, Candida kefyr, Candida krusei, Candida lipoly-
the diagnosis of probable disease. During the intervening pe-
tica, Candida lusitaniae, Candida parapsilosis, Candida rugosa,
riod, the patient had received appropriate antifungal therapy.
and Candida tropicalis), Cryptococcus neoformans, Rhodotorula
PCR positivity rates were higher in the proven and probable
rubra, Saccharomyces cerevisiae, and F. solani were not detected
IA categories and correlated well with radiological markers—
by the assay.
10 of the 12 patients with PCR-positive, proven/probable IA
Aspergillus PCR sensitivity and speci?city with clinical
had CT scans of the chest suggestive of possible fungal infection.
specimens.
There were 14 cases (11 with acute leukemia and/
Furthermore, 32 of the 33 patients whose CT scan ?ndings
or receipt of allogeneic stem cell transplantation [AL/SCT]) of
were negative also had negative PCR results. In total, 75 patients
proven/probable IA, 40 cases [13 AL/SCT] of possible IA, and
underwent a CT scan: 18 scans were suggestive of an invasive
149 cases (109 AL/SCT) classi?ed as at-risk for IA, generating
fungal infection, 24 were nonspeci?c, and 33 were negative.
an incidence rate of 6.9% (9.2% in the AL/SCT population).
The positivity rate for patients with proven/probable IA with
A total of 401 whole blood samples collected in EDTA vacu-
serial positive PCR results was 92.3%, compared with 15.0%
tainers from 203 patients (range per patient, 1–14 samples)
for those with possible IA (difference, 77.3%; 95% CI, 48.1%–
(table 1) were tested by PCR, with 51 patients having at least
87.5%) and 5.4% for at-risk groups (difference, 86.9%; 95%
1 specimen test positive by Aspergillus PCR, but, more signif-
CI, 60.9%–93.8%) (table 3). The sample positivity rate by use
icantly, 26 patients (12.8%) had serial positive results (range,
of serial positives for proven/probable cases was 60.0%, where-
2–10 positive results). Only patients with serial positive results
as, for possible and at-risk cases, it was 14.5% (difference,
were considered to be PCR positive. Single positive PCR results
45.5%; 95% CI, 31.6%–57.4%) and 10.7% (difference, 49.3%;
among multiple samples were considered false-positive results,
95% CI, 36.6%–64.0%), respectively (table 3).
and single positive PCR results for a specimen with no other
The statistical performance of the assay was calculated using
482 • CID 2006:42 (15 February) • White et al.

---

Figure 1.
The detection limits of the single-round Aspergillus PCR using colony-forming units (cfu) (A) or plasmid copy number (B)
different de?nitions of disease status [7]. The sensitivity, spec-
By use of serial positive PCR results, only a slight increase
i?city, positive predictive value, and negative predictive value
in sensitivity (92.3%) and a slight decrease in speci?city (94.6%)
of the Aspergillus PCR are shown in table 4.
were observed when compared with the original method [12].
However, a depreciation of 40% was seen for the positive pre-
DISCUSSION
dictive value by use of the Light Cycler–based assay. This de-
The value of PCR for diagnosing invasive fungal infections has
crease can be explained by the 8 patients with serial positive
yet to be determined, and PCR results are not included in the
PCR results who lacked EORTC-MSG clinical or microbiolog-
current EORTC-MSG criteria for de?ning invasive fungal in-
ical criteria for IA and were classi?ed as at-risk. Of a median
fection. Although PCR methods may lack standardization and
of 5 samples tested per patient (range, 3–8 samples tested per
optimization across centers, antigen detection provides varied
patient), the median number of positive samples per patient
performance in terms of sensitivity (29%–100% [19]). It is
was 3. In 4 patients, the positive PCR results spanned up to 4
included in the consensus criteria because of its commercial
consecutive weeks, during which the patients remained febrile
availability and good speci?city (185% [19]), despite a lack of
and neutropenic, with no other cause for their fever. Three of
agreement regarding the correct cutoff for positivity.
these patients received antifungal therapy, with 2 patients be-
Aspergillus PCR • CID 2006:42 (15 February) • 483

---

IA receive a de?nite diagnosis [21], which emphasizes the need
for improved diagnostic methods [20, 22].
Analysis of specimen positivity rate showed that 60%, 14.5%,
and 10.7% of samples for proven/probable, possible, and at-
risk cases, respectively, were positive by PCR. The 95% CIs
con?rm these differences and suggest that, 95% of the time,
the sample positivity rate for proven/probable cases would be
37%–64% greater than that for at-risk cases.
Since at-risk cases have low positivity rates for samples, serial
positive PCR results during 1 particular episode should be con-
sidered indicative of probable/possible IA. Applying these cri-
teria would mean that 7 of the 8 patients classi?ed as at-risk
who had serial Aspergillus-positive PCR results could be re-
garded as having possible IA, and it would enhance speci?city
and positive predictive values to 99.3% and 92.2%, respectively.
Figure 2.
Standard curve for the Aspergillus PCR, where Y p ?1.5705ln(X)
The high sensitivity of the assay has implications for initi-
+ 42.71. Both the correlation coef?cient (r ) and the correlation coef?cient
ation and continuation of antifungal therapy. The excellent
squared (R 2) were 1.00. The mean (?SD) crossing-point for 5 input copies
was 38.66 (? 0.78) cycles. The maximum variation in crossing-point for
negative predictive value (99.3%) will allow clinicians to with-
5 input copies was 1.44 cycles (3.7%).
draw therapy in cases that have responded. No patients in this
study received a diagnosis of other ?lamentous fungal infec-
coming PCR negative; no follow-up specimens were available
tions; however, the number of cases of non-Aspergillus ?la-
from the third patient. All 3 patients who received antifungal
mentous fungal infections is rising [23], and other assays [20]
therapy survived. The fourth patient received no antifungal
are needed to avoid false-negative results of non-Aspergillus
therapy and died, although autopsy was not performed.
infections or to withhold therapy.
Jordanides et al. [20] documented the problem of serial pos-
The optimal specimen for the molecular detection of IA is
itive PCR results in patients and commented on the dif?culty
still not resolved. Some groups prefer whole blood samples
in determining an actual “false-positive” result from an early
[8, 15], others prefer serum samples [24, 25], and some prefer
“true-positive” result, re?ecting the fact that PCR is more sen-
BAL samples [4, 26]. Serum has been reported to be an ap-
sitive than current diagnostic procedures. In clinical practice,
propriate sample for the diagnosis of IA [24]. However, only
other adjunctive procedures required for consensus diagnostic
free-circulating Aspergillus DNA will be targeted, because hy-
criteria usually are not performed or are performed at the
phae or phagocytosed intracellular fungal fragments will be
wrong time [3]. It seems inappropriate to use methods that are
removed by clot formation and centrifugation.
rarely used antemortem (histopathology), have transient results
If fungal fragments are present in blood, they should be
(CT), or are less sensitive (blood culture) as comparison in an
detectable by blood culture methods [24], particularly if lytic
assessment of the performance of a PCR method, unless a
systems are used [27, 28]. As few as 1–10 cfu of Aspergillus
structured regimen of testing is in place for all methods. This
conidia inoculated into blood culture bottles can be detected
highlights the dif?culties in determining a proven/probable case
within 24 h [29], but, in the clinical scenario, blood cultures
of IA, and it is not unexpected that only 6% of patients with
are rarely positive for Aspergillus (possibly because phagocytosis
Table 3.
Correlation between detection of Aspergillus by PCR and the level of con?dence in invasive
aspergillosis (IA) infection.
No. of patients
with positive PCR
No. of
Single
Serial
Patient
samples tested
Sample
positive
positive
positivity rate, %
(no. PCR
positivity rate, %
Classi?cation
result
results
(95% CI)a
positive)
(95% CI)a
Proven/probable IA (
)b
n p 13
0
12
92.3
70 (42)
60.0
Possible IA (n p
)
40
13
6
15.0 (48.1–87.5)
117 (30)
14.5 (31.6–57.4)
At-risk (n p
)
149
12
8
5.4 (60.9–93.8)
214 (35)
10.7 (36.6–64.0)
a Positivity rates are only for patients with serial positive PCR results, and 95% CIs were generated for differences between
proven/probable, possible, and at-risk categories.
b One patient was excluded.
484 • CID 2006:42 (15 February) • White et al.

---

Table 4.
Performance of Aspergillus PCR with use of different
References
de?nitions of disease status.
1. Marr KA, Carter RA, Boeckh M, Martin P, Corey L. Invasive asper-
gillosis in allogeneic stem cell transplant recipients: changes in epi-
Under invasive aspergillosis (IA) de?nitiona
demiology and risk factors. Blood 2002; 100:4358–66.
Value
A
B
C
2. Caillot D, Couaillier J-F, Bernard A, et al. Increasing volume and chang-
ing characteristics of invasive pulmonary aspergillosis on sequential
Sensitivity
92.3 (66.7–98.6)
34.0 (22.7–47.4)
92.3 (66.7–98.6)
thoracic computed tomography scans in patients with neutropenia. J
Speci?city
94.6 (89.8–97.3)
94.6 (89.8–97.3)
92.6 (88.0–95.5)
Clin Oncol 2001; 19:253–9.
PPV
60.0 (38.7–78.1)
69.2 (50.0–83.5)
46.2 (28.8–64.5)
3. Denning DW, Kibbler CC, Barnes RA. British Society for Medical
NPV
99.3 (96.1–99.9)
80.1 (73.6–85.3)
99.4 (96.9–99.9)
Mycology proposed standards of care for patients with invasive fungal
infections. Lancet Infect Dis 2003; 3:230–40.
NOTE.
Data are percentage (95% CI). Statistics were calculated using
4. Speiss B, Buchheidt D, Baust C, et al. Development of a LightCycler
serial positive-PCR results. NPV, negative predictive value; PPV, positive pre-
PCR assay for detection and quanti?cation of Aspergillus fumigatus
dictive value.
DNA in clinical specimens from neutropenic patients. J Clin Microbiol
a Under de?nition A, proven/probable cases are true positives, at-risk cases
2003; 41:1811–8.
are true negatives, and possible cases are excluded from statistical analysis;
5. Lo¨ef?er J, Henke N, Herbart H, et al. Quanti?cation of fungal DNA
under de?nition B, proven/probable/possible cases are true positives, and at-
by using ?uorescence resonance energy transfer and the Light Cycler
risk cases are true negatives; under de?nition C, proven/probable cases are
system. J Clin Microbiol 2000; 38:586–90.
true positives, and possible/at-risk cases are true negatives.
6. Pham AS, Tarrand JT, May GS, Lee MS, Kontoyiannis DP, Han XY.
Diagnosis of invasive mold infection by real-time quantitative PCR.
of fungal fragments by macrophages renders the fungal com-
Am J Clin Pathol 2003; 119:38–44.
7. Kawazu M, Kanda Y, Nannya Y, et al. Prospective comparison of the
ponent nonviable), whereas DNA can still be detected [30, 31].
diagnostic potential of real-time PCR, double-sandwich enzyme-linked
Detection of A. fumigatus in a murine model with intravenously
immunosorbent assay for galactomannan and a (1r3)-b-d-glucan test
induced fungemia 3 days after infection was 0% by blood cul-
in weekly screening for invasive aspergillosis in patients with haema-
ture and 100% by PCR [30].
tological disorders. J Clin Microbiol 2004; 42:2733–41.
8. Kami M, Fukui T, Ogawa S, et al. Use of real-time PCR on blood
Whole blood inoculated with DNA extracted from known
samples for diagnosis of invasive aspergillosis. Clin Infect Dis 2001;
amounts of A. fumigatus (10–1000 cfu) was reextracted and
33:1504–12.
tested by both single-round and nested Aspergillus PCR meth-
9. Lo¨ef?er J, Schimdt K, Herbart H, Schumacher U, Einsele H. Automated
extraction of genomic DNA from medically important yeast species
ods, but no samples tested positive (results not shown). This
and ?lamentous fungi by using the MagNA Pure LC system. J Clin
suggests that the procedure is not suitable for extracting free
Microbiol 2002; 40:2240–3.
DNA or that the DNA is degraded by DNAses present in the
10. Lo¨ef?er J, Herbart H, Schumacher U, Reitze H, Einsele H. Comparison
of different methods for extraction of DNA of fungal pathogens from
blood. Results for this method correlate with proven/probable
cultures and blood. J Clin Microbiol 1997; 35:3311–2.
cases of IA. Detecting DNA from nonviable or dead Aspergillus
11. Einsele H, Herbart H, Roller G, et al. Detection and identi?cation of
hyphae circulating in the blood is a possible explanation.
fungal pathogens in blood by using molecular probes. J Clin Microbiol
BAL samples are reported to give higher sensitivities and
1997; 35:1353–60.
12. Williamson ECM, Leeming JP, Palmer HP, et al. Diagnosis of invasive
speci?cities than blood samples [32]. Although PCR sensitivity
aspergillosis in bone marrow transplant recipients by polymerase chain
may be improved by testing BAL samples, improved speci?city
reaction. Br J Haematol 2000; 108:132–9.
must be questioned. Indeed, 25% of BAL samples from healthy
13. Yamakami Y, Hashimoto A, Yamagata E, et al. Evaluation of PCR for
detection of DNA speci?c for Aspergillus species in sera of patients
donors are PCR positive through inhalation of airborne As-
with various forms of pulmonary aspergillosis. J Clin Microbiol 1998;
pergillus spores [33]. Environmental contamination of blood
36:3619–23.
specimens is minimal, and blood is easier to obtain for initial
14. White PL, Archer AE, Barnes RA. A comparison of non-culture based
methods for the diagnosis of systemic fungal infections with an emphasis
and repeated sampling [21].
on invasive Candida infections. J Clin Microbiol 2005; 43:2181–7.
In summary, we describe the evolution of a manual DNA-
15. Lo¨ef?er J, Herbart H, Brauchle U, Schumacher U, Einsele H. Com-
extraction, nested-PCR method into a semiautomated real-time
parison between plasma and whole blood for the detection of Asper-
PCR assay that has improved sensitivity without a major re-
gillus DNA by PCR. J Clin Microbiol 2000; 38:3830–3.
16. Buchheidt D, Baust C, Skladny H, et al. Detection of Aspergillus species
duction in speci?city. Like other groups [34], we believe that
in blood and bronchoalveolar lavage samples from immunocompro-
if a patient has both PCR and galactomannan ELISA positive
mised patients by means of 2-step polymerase chain reaction: clinical
results, then it is likely that the patient has fungal disease, and
results. Clin Infect Dis 2001; 33:428–35.
17. Ascioglu S, Rex JH, de Pauw B, et al. De?ning opportunistic invasive
we believe that both tests should be used to screen at-risk
fungal infections in immunocompromised patients with cancer and
patients, with the concurrent use of both assays allowing con-
hematopoietic stem cell transplants: an international consensus. Clin
?rmation of diagnosis.
Infect Dis 2002; 34:7–14.
18. White PL, Shetty A, Barnes RA. Detection of seven Candida species
Acknowledgments
using the Light-Cycler system. J Med Microbiol 2003; 52:229–38.
19. Mennink-Kersten MASH, Donnelly JP, Verweij PE. Detection of cir-
Financial support.
We thank P?zer UK for their continued ?nancial
culating galactomannan for the diagnosis and management of invasive
support of this research.
aspergillosis. Lancet Infect Dis 2004; 4:349–57.
Potential con?icts of interest.
All authors: no con?icts.
20. Jordanides NE, Allan EK, McLintock LA, et al. A prospective study of
Aspergillus PCR • CID 2006:42 (15 February) • 485

---

real-time panfungal PCR for the early diagnosis of invasive fungal
MYCO/F Lytic bottle to the isolator tube, BACTEC Plus Aerobic F/
infection in haemato-oncology patients. Bone Marrow Transplant 2005;
bottle, and BACTEC Anaerobic Lytic/10 bottle and comparison of the
35:389–95.
BACTEC Plus Aerobic F/ bottle to the isolator tube for recovery of
21. Latge JP. Aspergillus fumigatus and aspergillosis. Clin Microbiol Rev
bacteria, mycobacteria, and fungi from blood. J Clin Microbiol 2001;39:
1999; 12:310–50.
4380–6.
22. Safdar A, Singhal S, Mehta J. Clinical signi?cance of non-Candida
29. Simonneau E, Kelly M, Labbe AC, Roy J, Laverdierem M. What is the
fungal blood isolation in patients undergoing high-risk allogeneic he-
clinical signi?cance of positive blood cultures with Aspergillus sp. in
matopoietic stem cell transplantation (1993–2001). Cancer 2004; 100:
hematopoietic stem cell transplant recipients? A 23 year experience.
2456–61.
Bone Marrow Transplant 2005; 35:303–6.
23. Marr KA, Carter RA, Crippa F, Wald A, Corey L. Epidemiology and
30. Hummel M, Baust C, Kretschmar M, et al. Detection of Aspergillus
outcome of mould infections in hematopoietic stem cell transplant
DNA by a nested PCR assay is superior to blood culture in an ex-
recipients. Clin Infect Dis 2002; 34:909–17.
perimental murine model of invasive aspergillosis. J Med Microbiol
24. Costa C, Costa J-M, Desterke C, Botterel F, Cordonnier C, Bretagne
2004; 53:803–6.
S. Real-time PCR coupled with automated DNA extraction and de-
31. Latge JP. The pathology of Aspergillus fumigatus. Trends Microbiol
tection of galactomannan antigen in serum by enzyme-linked im-
2001; 9:382–9.
munosorbent assay for diagnosis of invasive aspergillosis. J Clin Mi-
32. Buchheidt D, Baust C, Skladny H, Baldus M, Brauninger S, Hehlmann
crobiol 2002; 40:2224–7.
R. Clinical evaluation of polymerase chain reaction assay to detect
25. Kawamura S, Maesaki S, Noda T. Comparison between PCR and de-
Aspergillus species in bronchoalveolar lavage samples of neutropenic
tection of antigen in sera for diagnosis of pulmonary aspergillosis. J
Clin Microbiol 1999; 37:218–20.
patients. Br J Haematol 2002; 116:803–11.
26. Rantakokko-Jalava K, Laaksonen S, Issakainen J, et al. Semi-quanti-
33. Bart-Delabesse E, Marmarot-Khuong A, Costa JM, Dubreuil-Lemaire
tative detection by real-time PCR of Aspergillus fumigatus in bron-
ML, Bretagne S. Detection of Aspergillus DNA in bronchoalveolar lav-
choalveolar lavage ?uids and tissue biopsy specimens from patients
age ?uid of AIDS patients by polymerase chain reaction. Eur J Clin
with invasive aspergillosis. J Clin Microbiol 2003; 41:4304–11.
Microbiol Infect Dis 1997; 16:24–5.
27. Fuller DD, Davis TE Jr, Denys GA, York MK. Evaluation of BACTEC
34. Bretagne S, Costa J-M, Bart-Delabesse E, Dhedin N, Rieux C, Cor-
MYCO/F lytic medium for recovery of mycobacteria, fungi and bacteria
donnier C. Comparison of serum galactomannan antigen detection
from blood. J Clin Microbiol 2001; 39:2933–6.
and competitive polymerase chain reaction for diagnosing invasive
28. Vetter E, Torgerson C, Feuker A, et al. Comparison of the BACTEC/
aspergillosis. Clin Infect Dis 1998; 26:1407–12.
486 • CID 2006:42 (15 February) • White et al.

---